Introduction
Nitrous oxide (N2O) is a potent greenhouse gas that is
produced as an intermediate product of biological nitrogen
conversions in soils (Stevens et al., 1997). Denitrification is
the stepwise anaerobic reduction of aqueous nitrate
(NO3-) to nitrite (NO2-) and into the gaseous
forms N2O and benign dinitrogen (N2). It is the
major global contributor to N2O production in grassland
soils (Saggar et al., 2013) and is responsible for a significant
fraction of agricultural greenhouse gas emissions (IPCC,
2014). Denitrification is mediated by the action of four enzymes:
NO3- reductase (NAR), NO2- reductase (NIR),
nitric oxide (NO) reductase (NOR), and N2O reductase
(N2OR) (Zumft, 1997), which are encoded by the
nar/nap, nir, nor, and
nos genes, respectively. Taxonomically diverse bacteria,
archaea (Philippot et al., 2007; Tiedje, 1994; Ishii et al., 2010),
and eukaryotes (Zumft, 1997) are known to harbour two or more
denitrification enzymes. Denitrifying bacteria are particularly
widely distributed in pasture soils (Graham et al., 2014), and more
than 60 genera have been identified (Chen et al.,
2012). Denitrifiers with all four reductases are capable of
emitting N2 and are said to be “complete”
denitrifiers. Those denitrifiers that lack N2OR and emit
N2O as the final product of denitrification are called
“incomplete” denitrifiers.
NirS, nirK, and nosZ genes have been targeted
as functional markers of both complete and incomplete denitrifiers
in soils (Stres et al., 2008; Throbäck et al., 2004; Morales
et al., 2010; Enwall et al., 2010). The balance of complete and
incomplete denitrifiers in soils can determine the ratio of
N2O : N2 produced during denitrification (Philippot
et al., 2011; Bakken et al., 2012). Thus, denitrifier community
structure and abundance can be important factors in determining
nitrogen (N) loss and agricultural greenhouse gas emissions from
soils. Indeed, a recent synthesis of 82 datasets relating
bacterial community structure and environmental characteristics to
a variety of carbon (C) and nitrogen (N) cycling processes found
that microbial community structure data improved the power of
models to explain denitrification process rates better than for
any other ecosystem process (Graham et al., 2016). Still, strong
relationships are not always observed between denitrification
rates and denitrifier community structure and abundance
(Cavigelli and Robertson, 2000, 2001; Chèneby et al., 1998;
Mergel et al., 2001). In particular, the structure of denitrifier
communities in environmental samples is often poorly correlated
with soil or environmental factors that are known to influence
process rates (Dandie et al., 2011; Enwall et al., 2010; Philippot
et al., 2009), indicating that our understanding of the factors
controlling the diversity and function of denitrifying communities
is still inadequate. Moreover, there is a need to identify the
soil conditions in which the presence and activity of denitrifiers
are likely to lead to substantial N2O emissions so that
appropriate strategies for targeted and effective management can
be deployed or developed where they are lacking.
Pastoral dairy farming is a preferred land use in mild and wet
climates on relatively fertile soils and flat sites that occupy
low-lying positions in the landscape, as these locations support
high rates of pasture production (Saggar et al., 2013). The
combination of periodically anoxic soil conditions, high
concentrations of N in cattle excrement patches, and relatively
high microbial biomass at these sites combine to favour
denitrification as a major oxidative metabolic pathway. Despite
this, denitrification rates and potentials as well as N2O
emissions through denitrification vary widely among pasture soils
(Cayuela et al., 2013; Giltrap et al., 2012; Groffman et al.,
2006). Soil management practices, including the addition of organic
amendments such as plant residues, compost, manure, or effluent
irrigation, can increase soil fertility and microbial biomass and
may lead to structural shifts in soil microbial communities, which
in turn alter soil biochemical processes (Kennedy and Smith,
1995). The addition of crop residues to soils increases the
abundance of denitrifier genes and leads to greater
denitrification in soils (Barrett et al., 2016; Gao et al., 2016;
Henderson et al., 2010). Likewise, increasing soil water content
is associated with increasing denitrifier gene abundances in soils
(Liu et al., 2012; Mergel et al., 2001). Management practices that
alter the size of the denitrifier community in soils are also
likely to affect its denitrification enzyme activity (DEA), as the
abundance of denitrifier genes can be a strong determinant of DEA
(Čuhel et al., 2010; Deslippe et al., 2014; Enwall et al.,
2010; Hallin et al., 2009). However, the geologic origins of
a soil can determine its dominant properties over a range of soil
C and water contents (Bronick and Lal, 2005). Indeed, we
previously found that soil texture, drainage class, and latitude
were powerful regulators of denitrification end products
(N2 vs. N2O) and total emissions. Also, both the
forms and quantities of gases emitted could be predicted by the 16S
rRNA gene communities of soil samples (Morales et al.,
2015). However, we still lack detailed knowledge of how variation
in soil properties affects denitrifier populations and
denitrification. Better information on the role of soil
physicochemical characteristics in determining the size and
activity of denitrifiers may allow for improved and soil-specific
management of N2O emissions from pastoral agriculture.
Here, we sought a better understanding of the relationships
between the structure, abundance, and activity of denitrifiers
over a range of New Zealand dairy pasture soils, which varied
widely in soil properties and had different management
conditions. We investigated whether the properties of these soils drove unique
denitrifier communities that supported different DEA or were
likely to generate different N2O emissions. We expected to
find that the size and structure of denitrifier communities would
vary most strongly in accordance with soil water content and that
soil physical properties or management practices that increase
soil water would enhance the size and activity of denitrifiers.
Description of soils.
Soil
Location of the dairy farm
Geographical location
Soil abbreviation
Soil classification
Mineralogy class
Te Kowhai silt loam
AgResearch Ruakura, Waikato
37∘44′57.55′′ S 175∘10′27.06′′ E
TeK
Typic orthic gley
Glassy volcanic, kaolinitic
Otorohanga silt loam
Tokanui, Waikato
38∘11′19.70′′ S 175∘12′35.67′′ E
OH
Typic orthic
Allophanic
Horotiu silt loam
AgResearch Ruakura, Waikato
37∘46′30.80′′ S 175∘18′23.27′′ E
HR
Typic orthic allophanic
Allophanic
Tokomaru silt loam
Massey University, Palmerston North
40∘22′58.50′′ S 175∘36′31.01′′ E
TM
Argillic-fragic perch-gley pallic
Vermiculitic
Manawatu fine sandy loam
Longburn, Palmerston North
40∘22′56.99′′ S 175∘32′24.49′′ E
MW
Weathered fluvial recent
Illitic
Manawatu fine sandy loam(effluent irrigated)
Longburn, Palmerston North
40∘22′58.26′′ S 175∘32′21.65′′ E
MWEI
Weathered fluvial recent
Illitic
Paparua silt loam (Springston)
Springston, Christchurch
43∘38′15.97′′ S 172∘28′13.81′′ E
PS
Weathered orthic recent
Illitic
Paparua silt loam (Lincoln)
Lincoln, Christchurch
43∘38′43.91′′ S 172∘25′21.86′′ E
PL
Weathered orthic recent
Illitic
Lismore stony silt loam
Ashburton, Canterbury
43∘53′17.44′′ S 171∘38′28.43′′ E
LM
Pallic orthic brown
Vermiculitic
Mayfield deep silt loam
Methven, Canterbury
43∘38′30.12′′ S 171∘43′47.28′′ E
MF
No data
No data
Adapted from Morales et al. (2015).
Materials and methods
Sites and soils
Our aim was to sample soils that would encompass the range of
physicochemical conditions that predominate on New Zealand dairy
farms. We therefore targeted soils on the basis of their
geographical location (North or South Island of New Zealand) and
mineralogy (allophanic or non-allophanic soils). As soil water
content is a key factor affecting the structure and activity of
soil denitrifier communities (Liu et al., 2012; Mergel et al.,
2001), it was also important to sample in both wet and dry
seasons. We therefore sampled soils over a 6-month period from
winter to summer. Soil textures varied from a stony silt loam to
a fine sandy loam, and the sites ranged from poorly drained to
well drained (Table 1). We sampled soils expected to have the
greatest soil water contents in winter and those we expected to be
driest in summer, with other soils sampled between these times
(see the Supplement Table S1 for soil sampling dates). We collected soils from 10
different commercial dairy farms (Fig. S1 in the Supplement). All were fenced from
livestock and none had been grazed within 8 weeks of sampling.
All sites were dominated by perennial ryegrass (Lolium perenne) and white clover (Trifolium repens). Fertilisation regimes varied among the farms and
consisted of applications of 150–200 kgNha-1
annually. Detailed descriptions of the individual fertiliser
applications at the 10 farms are described in the Supplement.
Sampling and analysis of soil properties
At each farm, we randomly selected six blocks of 100 m2
for the collection of soil samples. At randomly selected locations
within each block, 25 soil cores (25 mm diameter × 100 mm long) were obtained using a steel
corer. The 25 cores from each block were pooled to form a single
composited sample per block (n=6 composited soil samples per
farm). All soil samples were collected between August and
December 2010 once from each site. Soil samples were taken to the
laboratory, individually homogenised, sieved to 2 mm, and
stored at 4 ∘C in plastic bags (10
sites × 6 replicates =60 samples). A subsample of
each soil was stored at -20 ∘C for molecular
analysis. We measured pH, nitrate–N (NO3-) and ammonium–N (NH4+)
(mineral–N), total nitrogen (TN), total carbon
(TC), Olsen phosphorus (P), and soluble C on the field-moist
sieved soils using standard protocols (for details see Jha,
2015). All soils were analysed for these parameters within 2 weeks
of sampling.
DEA was determined using the acetylene inhibition method described
in Luo et al. (1999), with the exception that we added
chloramphenicol to inhibit the de novo synthesis of
enzymes. Thus the values we report represent only the existing
enzyme activity in soils. DEA was assessed for all soil samples
within 2 days of collection. DEA incubation conditions and the gas
sampling methods are described elsewhere (Jha, 2015). We intended
to measure microbial biomass carbon (MBC) within 48 h of soil
sampling, but a technical problem with our set-up delayed
measurements of MBC for nearly 3 months. To standardise this
effect across soil samples we monitored changes in the size of the
MBC pool in two soils over 7 months. We found that no significant
changes in MBC occurred between 3 and 7 months for soils stored at
4 ∘C (see Table S1). We therefore report MBC data for all
soils that were stored at 4 ∘C for 4 to 6 months.
DNA extraction from soils
Within 6 months of soil sample collection, soil samples were
thawed on ice and a 0.25 g aliquot was obtained. DNA was
extracted from these soil samples using the Mobio
PowerSoil™ DNA Isolation Kit (Mobio,
Solana Beach, CA, USA) following the manufacturer
instructions. The yield and quality of DNA extracts were verified
as described in Deslippe et al. (2014). DNA was stored at
-20 ∘C until analysed.
Terminal restriction fragment length polymorphism (T-RFLP)
of denitrifier genes
Terminal restriction fragment length polymorphism (T-RFLP) was
performed to analyse the community structure and diversity of
nir and nos genes in soil samples. T-RFLP for
nirS and nosZ genes was conducted as described in
Deslippe et al. (2014) except that PCR for nir genes
occurred in a total volume of 25 µL reaction mixture,
which contained 2.5 µL of 10 × PCR buffer
(1 mMMgCl2), 0.5 mMMgCl2, 0.2 mM of each
deoxynucleotide triphosphate (dNTP), 1.25 U of Taq
polymerase (Thermofisher
Scientific®),
0.8 mgmL-1 bovine serum albumin (BSA),
1.0 µM of each primer, and 10 ng DNA template
per reaction. The PCR amplification consisted of an initial
denaturation of the DNA template at 94 ∘C for
30 s followed by 35 cycles of 20 s at
94 ∘C, 20 s at 56 ∘C, and 20 s at
68 ∘C. The reaction was completed by 10 min at
68 ∘C.
For T-RFLP of the nirK gene we used the primers Copper 583F and
909R (Dandie et al., 2011). The amplifications of nirK and
nosZ genes were achieved under slightly different conditions
than the nirS gene according to the specifications of the
reagents used for PCR. The PCR amplification was performed in
a total volume of 25 µL reaction mixture containing
10 µL of 2 × NEB Taq master mixes (New England
Biolabs® Inc.), 0.4 µM of
each primer, and 10 ng DNA template per reaction. PCR
consisted of an initial denaturation of the DNA template at
94 ∘C for 2 min followed by 35 cycles of
30 s at 94 ∘C, 1 min at 56 ∘C,
and 1 min at 72 ∘C. The reaction was considered
complete after 10 min at 72 ∘C.
The T-RFLP profiles generated for the soil samples were analysed
using Peak Scanner® v1.0 software (Life Technologies)
and as described in Deslippe et al. (2014). The total number of
terminal restriction fragments (T-RFs) per electropherogramme was
taken to indicate genotype richness per sample. We then calculated
the gene Shannon's diversity index and Pielou's evenness index
(Magurran, 1988) per sample and used one-way analysis of variance
(ANOVA) to determine if soils belonging to the three
physicochemical groups differed with respect to gene richness,
evenness, and diversity.
Quantitative polymerase chain reaction (qPCR) of total bacterial and denitrifier genes
Quantification of bacterial nirS, nirK, and
nosZ genes was accomplished using qPCR following the
methodology of Deslippe et al. (2014). Amplification efficiencies
of qPCR reactions for samples were within the range of values
(E=90–110 %) published previously (McPherson and Moller,
2006). The reactions were linear over 7 orders of magnitude and
sensitive down to 102 copies. The ratio of abundances of
denitrifier genes in environmental samples has been interpreted
previously as an index of the potential for complete
denitrification (Philippot et al., 2009). Here, we calculated the
nosZ : (nirS + nirK) of soil samples.
Statistical analysis
The normality and homoscedasticity of all soil physicochemical,
gaseous emission, and biological datasets were examined using
Anderson–Darling (Stephens, 1986) and Levene's tests,
respectively, in Minitab® 16 software (Minitab
Inc.). Box–Cox transformations (Box and Cox, 1964) were applied
to datasets as required to conform to model. The differences in
the means of soil characteristics, such as pH, nitrate–N
(NO3-–N) and ammonium–N (NH4+–N) (mineral N), total
nitrogen (TN), total carbon (TC), Olsen phosphorus (P), microbial
biomass carbon (MBC), soluble C, DEA, number of gene T-RFs, and
gene copy numbers, were assessed using a one-way analysis of
variance (ANOVA) test with soil type as a factor. Tukey's
studentised range test at an α=0.05 significance level was
used post hoc to reveal significant differences among means. The
relationships among the soil chemical characteristics pH, nitrate–N
(NO3-–N) and ammonium–N (NH4+–N), TN, TC, Olsen P,
MBC, DEA, number of denitrifier gene T-RFs, and gene copy numbers
were determined using Pearson correlation analysis.
In order to reduce the dimensionality of the many correlated soil
physicochemical characteristics, we performed a principal component
analysis (PCA) and included % soil water content (SWC),
% SWC at field capacity (% FC SWC), pH, TN, TC, soluble
C, Olsen P, and nitrate–N (NO3-–N) and ammonium–N
(NH4+–N) as factors in the PCA. Soils grouped along the
first and second ordination axes. We used a multiple response
permutation procedure (MRPP) to assess the statistical
significance of these groupings. MRPP calculates the
chance-corrected within-group agreement (A), a measure of
within-group homogeneity compared with that expected by chance,
where A=1 corresponds to identical members within each given
group (maximum effect of factor) and where A≤0 corresponds
to within-group heterogeneity equal to or larger than that
expected by chance (no effect of factor; McCune and Medford,
1999). We also calculated Pearson correlations among soil
microbial characteristics and the ordination axes and plotted
those that were significantly correlated (τ > 0.2) with axis
1 and 2 as vectors on the PCA.
Analysis of the nirS, nirK, and nosZ community
structure was based on threshold normalised peak heights of T-RFs
from electropherogrammes (Deslippe et al., 2014). Non-metric
multidimensional scaling (NMS) ordinations were performed using
Bray–Curtis distance (Bray and Curtis, 1957) in the programme
PC-ORD (McCune and Mefford, 1999). In order to illustrate how the
structure of denitrifier communities varied with the
physicochemical characteristic of soils, we calculated Kendall's
rank correlations among the physicochemical and biological
characteristics of soils with the NMS ordination axes in PC-ORD.
The significant correlates (τ > 0.2) were overlaid as vectors
on the NMS ordination plots.
Principal component analysis of the physicochemical
characteristics of soil samples collected at 10 dairy farms and
results of a multiple response permutation procedure (MRPP) to assess the
significance of soil origin and group. Vectors indicate significant
correlates (τ > 0.2) with ordination axes in the first and
second PC axes.
Discussion
Despite its relatively small total land area, New Zealand is
geologically diverse, and the 1.8 million hectares of land that
were managed as dairy pasture in 2015 (Dairy NZ, 2017) have soils
derived from a wide range of parent materials. Here we studied 10
dairy pasture soils that varied widely in texture, drainage class,
and management strategies. We found that the % FC SWC and
a gradient in mineral-N form accounted for the greatest variation
in soil physicochemical characteristics and that key microbial
parameters for denitrification, such as MBC and DEA, were
significantly positively correlated with higher soil
NO3-–N. In our study, these patterns were driven
primarily by only three soils: the two allophanic soils, which had
high % FC SWC (group 1), and the effluent-irrigated soil,
which had very high NO3-–N (group 2). The effluent
irrigated soil, which had the highest MBC, likely harboured
a larger population of nitrifiers with activities that generated the
NO3- required by denitrifiers and supported the
highest rates of DEA we observed. Nonetheless, our results are
consistent with previous reports that soil microbial biomass is
a key indicator of denitrification process rates (Drury et al.,
1991). From this perspective, across a wide range of soil
properties, the size of the MBC pool may be an important
coarse-scale indicator of soil N2O emissions under both
anoxic (denitrification) and oxic (nitrification) conditions.
Allophanic soils have high water content at field capacity, but
they adsorb copper and are therefore likely to select against
nirK denitrifiers, the periplasmic nitrite reductase of which
requires six copper atoms to maintain its trimeric structure. This
was reflected in our data by the very low richness, evenness, and
diversity of nirK T-RFs and in the very low numbers
of nirK gene copies relative to the other soils. We
expected this to also reduce the overall number of genes encoding
nitrite reductase in group 1 soils but did not observe
this. Instead we found that nirS denitrifiers replaced
nirK denitrifiers in allophanic soils so that the total
number of nir gene copies was equivalent to that in the
effluent-irrigated soil and significantly greater than the number
of nir copies in all other soils. Interestingly, despite
the large size of the nirS community, allophanic soils did
not, on average, have more diverse nirS communities than
other soils. However, the size and diversity of nirS
communities in allophanic soils was more variable than for other
soils. These findings suggest that allophanic soils support
relatively few microsites where denitrification driven by
nirS denitrifiers is the dominant respiratory pathway. New
Zealand's allophanic soils are porous and free draining with
relatively low bulk densities (Molloy, 1998). As such, anoxic
microsites conducive to denitrification are expected to be
few. Likewise, fewer active microsites for denitrification fits
with the low to moderate DEA we observed in the allophanic
soils. Allophanic soils are known to adsorb P (Hashizume and
Theng, 2007), and the binding of adenosine by allophanes may have
limited DEA in these soils despite their relatively large
nir populations. Nonetheless, we also found far fewer
copies of nosZ genes, relatively low nosZ
diversities, and the lowest nos : nir gene ratios in the
allophanic soils, suggesting that complete denitrifiers are
relatively rare in these soils. Consequently, where and when it
occurs, denitrification in allophanic soils is likely to lead to
significant N2O emissions. This result fits with other
work from our group, which indicates that allophanic soils emit
greater N2O : (N2O + N2) relative to other
soil types (McMillan et al., 2016). Taken together, these results
suggest that targeted management of nirS denitrifiers in
allophanic soils during wet seasons may be an effective strategy
to combat greenhouse gas emissions from pastoral agriculture in
volcanic regions.
The effluent-irrigated soil (MWEI), with physicochemical
properties that separated it from all other soils, was
characterised by very high NO3 and Olsen P concentrations,
relatively high pH (5.9), and high MBC, which supported very high
DEA. This moderately drained, fine sandy loam had the highest SWC
at the time of sampling. MWEI had the largest number of
nirK gene copies but only moderate numbers of nirS,
leading to intermediate total numbers of nir
genes. Likewise, it had the greatest diversity of nirK
genotypes but only moderate diversity of nirS
genotypes. These findings emphasise the potential for effluent
irrigation to increase denitrification enzyme activity, likely
through increasing both the size of the total microbial community
(MBC), SWC, and NO3 availability, which in turn selects for
denitrifiers. However, MWEI supported a significantly larger
population of nosZ denitrifiers than the other soils and
this led to the highest nos : nir of any soil. Overall,
these findings suggest that MWEI is likely to support a large and
active community of denitrifiers but that complete denitrification
may limit N2O emissions from this soil. Consequently,
management of greenhouse gas emissions from highly fertile pasture
soils like MWEI may benefit from strategies that limit NO3
availability in soils, such as the application of nitrification
inhibitors (e.g. DCD).
When considered in isolation, the seven soils of group 3 still
varied significantly with regard to physicochemical
characteristics. In particular MW, PS, and PL, which ranked higher
on axis 2 of the PCA, differed from the four remaining soils
(Fig. 1). For example, they had on average three times the
NO3 and half the NH4+ as the other group 3
soils. These soils also had the three highest pH values
(6.0–6.4), high MBC, and relatively low SWCs. They supported
relatively large numbers of nirS and nosZ
denitrifiers but only average numbers of nirK genes,
leading to overall intermediate nir : nos. Likewise,
these soils had intermediate diversities of nirS,
nirK, and nosZ genes and ordinations of the
nirS, nirK, and nosZ gene T-RFs failed to
distinguish these soils from those in the other groups. Despite
this, when incubated under non-limiting conditions, these three
soils together with MWEI supported the highest DEAs. These
findings indicate that denitrification responds quickly to SWC in
moderately fertile soils. Thus, careful management of
NO3- loads by limiting dairy stock, using nitrification inhibitors, or both is also likely to be useful
in
limiting greenhouse gas emissions from these soils during wetter
periods of the year.
The four remaining soils of group 3 were the least fertile and the
most acidic (pH 4.8–5.7) with the lowest MBC in our study. They
were also among the driest. Despite moderately high total numbers
of nirS, nirK, and nosZ genes, the
nos : nir ratios in these soils were equivalent to the other
soils of group 3 and intermediate overall. With the exception of
the two allophanic soils, the four remaining soils of group 3 had
the lowest DEA, which was on average about one-quarter of that
measured in the other group 3 soils. Taken together, these results
suggest lower risk of N2O emissions from these soils, as
oxic conditions and low concentrations of substrates are likely to
limit denitrification much of the time, and moderately high
numbers of nosZ denitrifiers will favour some complete
denitrification of this smaller total N pool.
Overall, ordinations of T-RFLP data revealed no structuring of the
nirS, nirK, or nosZ communities according to the
three groups that defined the major physicochemical
characteristics of the soils, a gradient in soil water content at
field capacity, and a gradient in mineral-N form. Rather, SWC at
the time of soil sample collection and Olsen P were the primary
drivers of the structure of denitrifier communities. Given the
overall high correlation of SWC and Olsen P in soil samples, this
result is likely to indicate considerable plasticity of the
denitrifier community in response to ambient soil water
content. In the wettest soils, we found strong and significant
positive correlations between the diversity of nirS,
nirK, and nosZ genes and DEA. We also found strong
and significant positive correlations between nirK and
nosZ gene copy numbers and DEA in those soils. However,
these relationships broke down for soils with moderate or low
SWCs. While the aim of our study was to sample pasture soils over
a wide range of physicochemical characteristics in order to gain
insight into the properties of the denitrifier communities they
support, seasonal variation in the structure of nirS and
nirK denitrifiers in cultivated and pasture soils is well
established (Wertz et al., 2016; Tatti et al., 2017; Bent et al.,
2016; Yu et al., 2016; Smith et al., 2010). The plasticity of the
denitrifier community in response to ambient soil water content
together with the strong correlations between the size, diversity,
and activity of denitrifying communities in very wet soils
suggests that future work towards characterising the denitrifier
communities most likely to contribute to greenhouse gas emissions
from pastoral soils should focus sampling efforts on the wettest
times of the year.
Several lines of evidence collected here suggest that nirK
denitrifiers were more sensitive to the range of physicochemical
characteristics in soils than were nirS denitrifiers. For
example, the gene copy numbers and diversity metrics of nirS
communities were fairly uniform across soil groups but varied
significantly for nirK denitrifiers. Likewise, the patterns
of nirK diversity across soil groups were mirrored by the
patterns of nos : nir, suggesting that changes in the
size of the nirK community had a dominant influence on the
overall ratio of complete and incomplete denitrifiers in soil
groups. Independent shifts in nirS and
nirK community structures in response to common
physicochemical characteristics were recently observed in
a eutrophic reservoir (Zhou et al., 2016). In contrast, the
structure of nosZ communities did not correspond to any
physicochemical property measured. Together, these results may
suggest that nirS and nosZ genotypes are
equivalently adapted to the physicochemical conditions of a wide
range of dairy pasture soils, while nirK denitrifiers are
more sensitive. Given that our data suggest greater N2O
emissions in allophanic soils where nirK denitrifiers are
few, it may be the microbial communities dominated by nirS
denitrifiers that should be the target of efforts to reduce greenhouse
gas emissions from pasture soils. However, further work is
necessary to confirm whether microbial communities dominated by
nirS denitrifiers support greater N2O emissions
than nirK denitrifier communities of equivalent size.